I think one of the excuses given is that the software grew in complexity and all the software updates pushed the chip harder than expected. (A fair excuse - I am not sure).
Uhmm... the emmc got worn out not because of the updates. Think about it, these things are rated in thousands of erase-program cycles. They could have updated the car every day for almost 2 decades (each update only updates half the chip) before reaching the rated cycles. Tesla did a horrible system design job by not considering the amount of logging their choice of software (Linux) has enabled by default. A number of hackers have told Tesla, the logging is unnecessary and just wears out a chip. So even when informed about their bad design choices (or lack thereof) they still didn't do anything about it, until MCU's started failing in large quantities. The secondary reason, which added to the emmc wear, was when Tesla added Autopilot and decided to do "fleet learning", i.e. collecting data (on the emmc storage) and uploading it to the mothership - this is why post 2014 MCU1 failed sooner than the earlier, non-AP cars. All signs point to a complete lack of design decisions - nobody looked / measured/ verified at how much data was being written, because nobody cared. "It worked on my desk, it worked for Elon on his commute this morning, it's done".
I totally agree that Tesla did not use automotive grade electronics. But what that a bad decision?
Absolutely, and not just hardware. Non-automotive grade software is just as bad. Tesla being clueless about the fact that you should budget things like emmc writes likely was caused by the fact that they were using a PC or tablet style operating system, where nobody cares about wearing out storage because they are expected to have 2-4 year useful lifespan. Btw, no other car manufacturer uses Linux for any driving functions - you will not be able to safety certify it (see
ISO26262)
I understand the eMMC issue has nothing to do with "grade" of course. They knew it had limited cycles but I suspect they calculated a lifespan and felt it wouldn't be an early failure point.
Calculated based on what data? It doesn't sound like they measured anything, or those calculation would have told them it would be an early failure. Perhaps someone asked "hey guys, do you think 3,000 writes is enough" and someone else answering "who cares, we'll be rich and retired by then".
Listen - Tesla messed up. I agree with you on that. You are bitter - and I get it. But I guess I feel like your bitterness is not always helpful. Have you thought of just moving on?
Would Tesla move on if I just stopped paying for the car, or I messed up and paid them for a new Plaid S with a check that bounced (perhaps I could claim the check was capable of the full amount, but limited by my account balance, just like my P85D was capable of 691hp, but limited by the battery)? I'll tell you what, if they give me a full refund on my car, I'll move on. Let's see they will.
Just to counter your experience. My 2015 has a clear screen and very limited service visits. Our 2019 M3 has never seen the service center. Anecdotes of course.
So like I've always been saying, Tesla is like a Hyundai of the 90's. I believe the Munro guy compared them to Kia of the 90's instead, but the same logic applies - when you buy one, you might lucky and get one which lasts without issues, or one which spends more time in the shop than on the road, or anything in between. Consumer reports in the 90's used to track your best chance of getting good Hyundai, depending on the day of the week it was manufactures, Tesla fans have similar metrics, but based on Elon's moods and end of quarter pushes.
My 2013 Leaf - mainstream manufacturer - had a 2G modem. They wanted $400 to get to 3G - back in 2018. I bet that 2G modem was automotive grade - lot of good that did.
Tesla modem upgrade (3G to LTE) ran me $500+tx ($550) in 2016, so the leaf was cheaper.
Does any other manufacturer have a $1500 price on main computer replacement?
The other guys don't have a "main computer", a.k.a. a single point of failure.
Lastly, "safety critical parts"? I mean I am happy that is the NHTSA's position but are we really such bad drivers that turn signal sounds and back up camera are critical? (Yes - window fogging is an issue - in certain specific situations a serious safety issue) But it isn't like the brakes stop working....
And it isn't like a HVAC fan blower isn't a wear part. As in, any car can have the HVAC fail and cause a safety concern. I doubt that is considered "critical".
The infotainment system controls more than that, depending on which week (yes, Tesla used to vary their design in production weekly). Some cars the blinkers don't work at all, not just sounds. Headlights and rear lights are controlled via MCU. Backup camera is now considered safety critical (which can be especially bad if dying MCU is freezing up and displaying the rear image with a delay). By the way, HVAC failure would be considered safety critical. If some model car out there is having massive HVAC failures within 8 year lifespan, and the HVAC part was not listed as required maintenance part, it would also be a recall. Remember, recall is when a lot of the cars of a particular vintage suffer the same issue earlier than expected.
Safety critical in automotive is not just RUD events, to borrow from Space X terminology.
